ABSTRACT. A ring-shea r device was used to study the factors that co ntrol the ultimate (steady) streng th of till at high shea r strains. Tests a t a steady strain ra te and at different stresses norm al to th e shearing direction yielded ultim ate friction angles of26.3° and 18.6° for tills conta ining 4% and 30% clay-sized particles, respectivcly. Other tests at steady normal stresses a nd va ri abl e shear-strain rates indi cated a tendency for both tills lO weaken slightly w ith increasing stra in rate. This weakening may be due to sma ll increases in till porosity.Th ese res ults provide no evidence o f v iscous behavior and suggest that a Coulombplas ti c idealization is reasonable [or till deformation. Howeve r, vi scous behavior has often been sugges ted on th e basis of di stributed shear stra in observed in subglaci a l till. We hyp ot hes ize that deform ation may becom e distributed in till th at is deformed cyc licall y in response to flu ctu ations in basal wa ter pressure. During a deform ation event, transient dil ation of discre te shea r zones should cause a reduction in intern al pore-water pressu re that should streng then th ese zo nes rela tive to the surro unding till, a process call ed dilarant harde ning. Consequent changes in hea r-zo ne positi on, when integrated over time, may yield the obse rved di tributed stra in.

[1] Field measurements of debris-bed friction on a smooth rock tablet at the bed of Engabreen, a hard-bedded, temperate glacier in northern Norway, indicated that basal ice containing 10% debris by volume exerted local shear traction of up to 500 kPa. The corresponding bulk friction coefficient between the dirty basal ice and the tablet was between 0.05 and 0.08. A model of friction in which nonrotating spherical rock particles are held in frictional contact with the bed by bed-normal ice flow can account for these measurements if the power law exponent for ice flowing past large clasts is 1. A small exponent (n < 2) is likely because stresses in ice are small and flow is transient. Numerical calculations of the bed-normal drag force on a sphere in contact with a flat bed using n = 1 show that this force can reach values several hundred times that on a sphere isolated from the bed, thus drastically increasing frictional resistance. Various estimates of basal friction are obtained from this model. For example, the shear traction at the bed of a glacier sliding at 20 m a À1 with a geothermally induced melt rate of 0.006 m a À1 and an effective pressure of 300 kPa can exceed 100 kPa. Debris-bed friction can therefore be a major component of sliding resistance, contradicting the common assumption that debris-bed friction is negligible.

Particle fabrics of basal tills may allow testing of the bed-deformation model of glacier flow, which requires high bed shear strains (> > > > >100). Field studies, however, have not yielded a systematic relationship between shear-strain magnitude and fabric development. To isolate this relationship four basal tills and viscous putty were sheared in a ring-shear device to strains as high as 714. Fabric was characterized within a zone of shear deformation using the long-axis orientations of fine-gravel and sand particles and the anisotropy of magnetic susceptibility (AMS) of small (~5·8 cm 3 ) intact samples. Results indicate that till particles rotate toward the plane of shearing with long-axis orientations that become tightly clustered in the direction of shear (0·78 < < < < < S 1 < < < < < 0·94 for three-dimensional data). These strong, steadystate fabrics are attained at shear strains of 7-30, with no evidence of fabric weakening with further strain, regardless of the specific till or particle-size fraction under consideration. These results do not support the Jeffery model of particle rotation, which correctly describes particle rotation in the viscous putty but not in the tills, owing to fluid-mechanical assumptions of the model that are violated in till. The sensitivity of fabric development to shearstrain magnitude indicates that, for most till units where shear-strain magnitude is poorly known, attributing fabric variations to spatial differences in other variables, such as till thickness or water content, will be inherently speculative. Attributing fabric characteristics to particular basal till facies is uncertain because shear-strain magnitude is unlikely to be closely correlated to till facies. Weak or spatially variable fabrics, in the absence of postdepositional disturbance or major deviations from unidirectional simple shear, indicate that till has not been pervasively sheared to the high strains required by the bed-deformation model. Strong flow-parallel fabrics are a necessary but insufficient criterion for confirming the model.

[1] Wet-based portions of ice sheets may move primarily by shearing their till beds, resulting in high sediment fluxes and the development of subglacial landforms. This model of glacier movement, which requires high bed shear strains, can be tested using till microstructural characteristics that evolve during till deformation. Here we examine the development of magnetic fabric using a ring shear device to deform two Wisconsinage basal tills to shear strains as high as 70. Hysteresis experiments and the dependence of magnetic susceptibility of these tills on temperature demonstrate that anisotropy of magnetic susceptibility (AMS) develops during shear due to the rotation of primarily magnetite particles that are silt sized or smaller. At moderate shear strains ($6-25), principal axes of maximum magnetic susceptibility develop a strong fabric (S 1 eignevalues of 0.83-0.96), without further strengthening at higher strains. During deformation, directions of maximum susceptibility cluster strongly in the direction of shear and plunge ''up-glacier,'' consistent with the behavior of pebbles and sand particles studied in earlier experiments. In contrast, the magnitude of AMS does not vary systematically with strain and is small relative to its variability among samples; this is because most magnetite grains are contained as inclusions in larger particles and hence do not align during shear. Although processes other than pervasive bed deformation may result in strong flow parallel fabrics, AMS fabrics provide a rapid and objective means of identifying basal tills that have not been sheared sufficiently to be compatible with the bed deformation model.